> I would read the official documentation of each db, forums, blog posts, stackoverflow entries, etc. It was time consuming on the searching side. The time it took to read all the sources was fine for me (it's learning time, so that's always welcomed).
This learning time that you welcomed is what you will now miss out on. The LLM gives you an answer, you don't know how good it is, you use it, and soon enough, if running into a similar issue, you will need to ask the LLM again since you were missing out on all that learning the first time which would have enabled you to internalize all the concepts.
It's like the regex engine example from the article. An LLM can create such a thing for you. You can read through it, it might even work, but the learning from this is orders of magnitudes less than what you get if you build this yourself.
I think it depends. LLMs can link references of where they took the content they throw at you. You can go and read such references. I like what LLMs provide, and at the same time I don't wanna blindly follow them, so I always allocate time for learning whether it's with LLMs or not.
You'd be surprised. A number of fairly technical people who are just not that familiar with ML I know got confused by this and believed the models were actually being tuned daily. I don't think that's universally understood at all.
That has actual practical implications and isn't just pedantry. People might like some model and avoid better dialog engines like perplexity believing they'd have to switch.
I think you either didn't read my response or missed the point. No matter if the LLM output is useful or not, the learning outcome is hugely impacted. Negatively.
It's like copying on your homework assignments. Looks like it gets the job done, but the point of a homework assignment is not the result you deliver, it's the process of creating that result which makes you learn something.
I mean, as with code, the actual typing is not really the bottleneck.
I also basically read the right pane rendered output, but mostly as a "reading out what I've written and evaluating whether it sounds good" most of the time, not really messing with layouting (especially that LaTeX and Typst does that very well, I can be reasonably sure that my paragraphs will have decent hypens and such).
I have written some documents for which I have yet to even look at the 40+ pages produced in the PDF as output. I'll admit thought that I wasted some time finding a nice template before starting to write.
And you didn't have the mental capacity to abstract from the colored balls to whatever application domain you were interested in? Does everything have to come pre-digested for students so they don't have to do their own thinking?
I had, and still have. The problem is, most people are exposed to this stuff way before they have even a single application domain they're even remotely interested in.
It's really the same problem as with math in school in general ("whatever is this even useful for?") - most people don't like doing abstract, self-contained puzzles with no apparent utility, but being high-stakes (you're being graded on it).
> It's really the same problem as with math in school in general ("whatever is this even useful for?")
That argument is a strawman whenever it comes up because it applies to every subject. High jump? Napoleon wars? Molar weight of helium? English literature in the 19th century? What is any of that ever "useful" for? To understand the world which you live in. What a lack of education leads to is blatently obvious with the current U.S. administration. It's not about each school lesson directly translating into monetary value in a later job, neither w.r.t colored balls nor with knowing how the american civil war started.
I see where you are coming from, and overtesting is a thing, but I really believe that the baseline of quality of all software out there is terrible. We are just so used to it and it's been normalized. But there is really no day going by during which I'm not annoyed by a bug that somebody with more attention to quality would have not let through.
It's not about space rocket type of rigor, but it's about a higher bar than the current state.
(Besides, Elon's rockets are failing left and right, in contrast to what NASA achieved in the 60s, so there are some lessons there too.)
I think there's a pretty big difference between QA (letting bugs go by) and A/B testing, and your post appears to me to be conflating the two. I would argue that you are better off spending your time QAing a feature that you have high confidence is positive ROI, than spending weeks waiting for an A/B test to reach stat sig.
I don't disagree with your statement, I just think you are addressing a different problem from A/B testing and statistical significance.
Perfect is the main enemy of good. I rather use Signal to escape the big tech clown show than wait for another decade for the perfect tech to come along, meanwhile either not communicating with anybody or using the exact services I really want to avoid.
I'm still waiting for the "other issues" to be explained that Signal supposedly has. I'm ok with my contacts knowing my phone number, and I opened the Signal account ages ago. Anything else to be concerned about?
I feel like "perfect is the enemy of good" only works when you (still) have to put considerable effort to make it better, e.g. when building software.
However currently there are already better alternatives than Signal, so in my personal opinion I feel like that saying does not apply.
It's very fine if you (and most people) are OK with sharing some personal information with a United States organization. That does not mean that everybody is fine with that, or that it's a very good solution to a chat service problem. I'm glad that Signal is a good match for your needs. But there are those of us who would rather see a decentralized service with which no personal information has to be shared.
In these kinds of discussions, I often find it a little strange when others decide that a certain solution or product must be good for everyone only because they are fine with it themselves.
Seems that phrase “perfect is the enemy of the good” is a relativistic argument. But the title’s frame is “ethics”, which one definition describes as “what is good in and of itself”. In that frame, perfection is the point, no? Though, I imagine you argue in this framework by elevating some aspects to that high standard, and work to convince other aspects are secondary. Otherwise, result is a preference argument where the trade offs you made are silent or obscured behind the practicality of your choices.
I have sympathies for this argument, though it boils down to trust. Even if you roll your own client, you still need to trust some things outside of your control, be it your build environment, your phone, whatever. But most people will use somebody else's client, so need to trust whoever built that one. Or whoever supposedly audited it. The Signal authors just play that role here. Their business model is fundamentally different that that of Google or Meta, which is a main source of trust people are putting into it. Offloading the exposure to a minimum (just the client which is open source) is another. Yes there are ways around all that for an attacker, but in the end it's a game of likelihood. A journalist or dissident fearing for their life may have a different conclusion than mom and dad who want to coordinate a birthday party without big tech reading those messages and selling them to ad companies. It's good to err on the side of caution, but acting like I am the former while in reality I'm closer to the latter user type is in the end just theatre.
The situation is not as black-and-white as you paint it. English is an ambiguous language. It's a "no op" in the sense that there is the possibility that they ship an app update in which the crypto is comprimised. Fair, that's possible, just like it's possible that your phone is backdoored or the CIA has installed hidden cameras in your bedroom. But as long as these things have not happened, specifically as long as Signal ships an app which corresponds to the code up on github which has been audited time and time again, it's not a no op and works perfectly fine. That's very different from sending a plain-text email or painting the message contents on your window. Please stop conflating these two.
> Signal's privacy terms were last updated in 2018. We are in 2025 now. It is unimaginable for any operational organization not to update their terms for 7 years.
As a privacy-concious user, I always get suspicious abouy privacy policy changes. They always become more loose instead of doing anything to my advantage. Typically because company has found a new way to use my data to make money and their lawyers realized that this requires relaxing the privacy policy. It's a good thing Signal is not playing that game.
Um, Whatsapp does that too?! Maybe there is a way to use it without a phone number, but the most common use is to have a phone number on some sign / store front with a Whatsapp symbol next to it. In many countries that's the default way to do business. Scolding Signal to use phone numbers is just weird in comparison.
This learning time that you welcomed is what you will now miss out on. The LLM gives you an answer, you don't know how good it is, you use it, and soon enough, if running into a similar issue, you will need to ask the LLM again since you were missing out on all that learning the first time which would have enabled you to internalize all the concepts.
It's like the regex engine example from the article. An LLM can create such a thing for you. You can read through it, it might even work, but the learning from this is orders of magnitudes less than what you get if you build this yourself.